How to evaluate MusicGen

facebookresearch / audiocraft

Audiocraft is a library for audio processing and generation with deep learning. It features the state-of-the-art EnCodec audio compressor / tokenizer, along with MusicGen, a simple and controllable music generation LM with textual and melodic conditioning.

MIT License

20.23k stars 2.03k forks source link

How to evaluate MusicGen #302

Open Hussain-Safwan opened 9 months ago

Hussain-Safwan commented 9 months ago

I've trained the model with my custom dataset and now want to analyse if it performs better on certain kinds of prompts. Please let me know,

how I can evaluate my finetuned model
which metric am I to use

piyuch commented 8 months ago

Hi @Hussain-Safwan , sounds great. How did you manage to train musicGen with your custom dataset ?

Did you use dora for training musicGen ?
If you used Dora, did you spin up slurm cluster. Is there are way to by pass the slurm cluster setup ?

thanks a lot in advance.